Project Summary The development of targeted therapy (BRAF/MEKi) and immune checkpoint blockade (ICB) targeting the co-inhibitory receptors CTLA-4 and PD-1 have revolutionized the treatment of metastatic melanoma. However, only a subset of patients maintain durable responses, and many people experience substantial side effects of therapy. Predicting therapeutic response in individual patients remains a critical and unresolved issue. Furthermore, the series of key genomic and epigenetic events driving progression and resistance to therapy is incompletely understood. The guiding hypothesis of this proposal is that (a) resistance to ICB and targeted therapy is mediated by tumor intrinsic and extrinsic mechanisms, some of which may be elucidated by systematic multi-modal molecular characterization of the tumor and tumor microenvironment; and (b) applying modern machine-learning and statistical approaches to molecular and clinical data from patient tumors will inform development of new therapeutic approaches and predictive models to improve patient care. Identifying and validating predictors of intrinsic resistance to BRAF/MEKi and ICB across large human cohorts has been limited to date. Aim 1 of this proposal applies genomic and transcriptomic characterization of pre-treatment tumors to large cohorts of patients treated with BRAF/MEKi, PD-1i, and CTLA-4i in order to discover and to validate molecular and clinical markers of response and resistance. Machine learning approaches will integrate these markers into parsimonious models predicting response. A differential analysis using mutual information will be conducted to reveal markers that predict differential response to therapy. A significant proportion of patients do not respond or maintained sustained responses to immunotherapy, and there is a critical need to characterize the acquisition or selection of drivers that confer resistance to immunotherapy. Aim 2 of this proposal develops algorithms using molecular characterization of longitudinally collected tumor samples across multiple anatomic sites to discover genomic and epigenetic drivers of progression and resistance to immunotherapy using phylogenetic analysis as the backbone of discovery. Finally, the ability to detect novel tumor driver mutations present at low frequencies is strongly dependent on cohort size. Aim 3 of this proposal leverages all genomically characterized melanomas to perform a meta- analysis using state-of-the-art and novel algorithms to discover novel driver mutations present at low frequencies with a focus on tumor subsets that lack known targetable drivers. These studies will expand the actionable landscape of genomic and epigenetic alterations in metastatic melanoma, advance our understanding of intrinsic and acquired resistance to targeted and immunotherapies in melanoma, and establish a framework to predict response in individual patients, which may impact patient care in melanoma and have applicability in other disease settings.